I know that my 400D's ISO range starts at 100, and I've heard the 5D goes as low as 50. But if you want to sit your camera on a tripod and do a long exposure of water flowing (for example), then in bright daylight you might need a neutral density filter.
But I'm wondering, why is that the camera can't just be set to ISO 25 or lower, thus removing the need for a filter for these kind of shots? My simplistic understanding is that ISO affects the gain applied to the photosensitivity of the pixels on the sensor. So what's to stop it being set very low?
Hope some technical expert can satisfy my curiosity here. Thanks.
I don't know the why, but Kodak's older pro bodies had this feature -- going down to iso 6, I think! I think this could be very useful, and I also wonder why it has not been pursued.
no tech expert here, but from what I understand, ISO 50 (ISO expansion "L") doesn't give any better image quality...in fact, a bit worse, if I remember correctly (unlike film - ah, the days of Ektar 25...showing my age here)
I agree it would be useful for desired low shutter, though.
It's outside the working limits of the sensors. When you go outside the working limits, you lose dynamic range. It would be nice if they had a built in 2 or 3 stop ND filter option though --- maybe have it slide across next to the shutter assembly.
I've wondered the same thing...but as an aside, I've been playing with trying to take multiple images (from a tripod, of course) to blend together to get the same effect as an ND filter for long exposures...Seems like it ought to work but so far I haven't had good success...but I think that is probably due to my limited Photoshop technique rather than the concept.
Another concept if the detectors can't go below ISO 100 type sensitivity is to mecahnically or electronically sample the sensor periodically over alonger exposure time...but that's the general thought behind what I've described above.
Edited by Dean DuBois on Oct 27, 2007 at 05:42 AM GMT
The basic circuit (sensor + amplifier) has a finite range. Putting more (passive) attenuation between the sensor and amplifier just drops the signal to noise (dynamic range). So the best way to handle this is to reduce the light outside of the system. Either stop down more or use an ND filter (or both if necessary).
The problem with ND filters is that you never know what the level of darkening you need will be...if you could make the adjustement in the camera or in post, you wouldn't have to worry about it...
Our Sony DNW9 cameras have a function which gives you -3db of gain, the equivalent of a third of a stop. After that the picture starts to look too cruddy. I think it can go further but our techos elected not to take that route. The problem is that the picture just starts to look pretty flat. I only ever use it to soften backgrounds but I don't do it very often.
The only alternatives after that are to change filters, throw in the doubler or increase shutter speed. All of those come with some trade offs.
Dean DuBois wrote:
I've wondered the same thing...but as an aside, I've been playing with trying to take multiple images (from a tripod, of course) to blend together to get the same effect as an ND filter for long exposures...Seems like it ought to work but so far I haven't had good success...but I think that is probably due to my limited Photoshop technique rather than the concept.
Try HDR software, I think that is what you are trying to do right? Increase the dynamic range of your images.
Think of the pixels as buckets that catch light.
They can catch a certain number of photons (about 50,000 ) before they are full. They go from "empty 0% reading" (black) to "full 100% reading" (white).
If you crank up the ISO (from e.g. 100 to 200 or 400 or 800), the on-board amplifier increases the read-out from a 1/2 full or 1/4 full or 1/8 full bucket and sets it to 100%.
The "base" ISO of Canon's cameras just "happens" to be around 100. In a scene with highlights that would be close to blowing out at an ISO of 100, the buckets would be , say 98% full. For the same scene set to ISO 50, you'd let in twice as many photons because the aperture would be the same but the shutterspeed would be twice as long, so many of the buckets begin to "overflow" - they can only go up to being 100% full - and more of the highlights would be blown. Therefore, as others have pointed out, you have to cut down on the photons to get longer exposures by using a filter...
A small point... ISO 50 in a 5D is really ISO 100 adjusted as you would do in post-processing with an overexposed image. In other words, it is a "fake" ISO 50...
htbyron wrote:
I don't know the why, but Kodak's older pro bodies had this feature -- going down to iso 6, I think! I think this could be very useful, and I also wonder why it has not been pursued.
But then they were rubbish at anything over about 100 ISO
10 years ago, most of films sold were 100 ISO and DSLR manufacturers didn't want to disturb the pros habits when introducing these brand new `D' products, thus the sensors were built based on a 100 ISO basis. The trend was (and is) more on the opposite side, gaining the highest ISO without losing any image quality ; that means gaining and counting accurately photons even when they're only a few (there is some quantum physics theory in between making the task very hard, recent progress is amazing!).
However, camera makers could rather easily (compared to high ISO) make sensors with higher capacities and be able to decrease the low ISO limit. It would seem they've got other fish to fry.
Low ISO is most often used when you want a long exposure. Say if you want a 10 seconds exposure and your camera meters the light to give 2.5 seconds at ISO 100, just take 4 exposures and average them in post processing. That is effectively the same as emptying the "bucket" and counting the photons 4 times during one exposure, to give you one ISO 25 exposure. This process could be done with in-camera processing as a low ISO mode.
alundeb wrote:
Low ISO is most often used when you want a long exposure. Say if you want a 10 seconds exposure and your camera meters the light to give 2.5 seconds at ISO 100, just take 4 exposures and average them in post processing. That is effectively the same as emptying the "bucket" and counting the photons 4 times during one exposure, to give you one ISO 25 exposure. This process could be done with in-camera processing as a low ISO mode.
There are probably engineering issues with arranging the details and most people would rather the effort go to the high end - as you said, it is something that does not require camera support.
Damian R wrote:
Hope some technical expert can satisfy my curiosity here. Thanks.
Put simply, when you change the "ISO" on your camera, you are doing nothing to the sensor! Changing the "ISO" is equivalent to changing the volume on your stereo. However, try to turn down the volume below the lowest point on the knob!
When you change the ISO on the 5D to 50, what ends up happening inside the camera electronics is that the digital value (after the analogue to digital conversion) gets divided by 2 (to make up for overexposing the image by a factor of 2 - yes "ISO 50" is really "ISO 100"). If you want an "ISO 25" then one would divide by 4, etc.
As was mentioned earlier, the problem with this approach is that once the sensor diode saturates you get no further information, so you end up with lots of blown (i.e., RGB values = 255) pixels. In other words, less headroom, and thus less DR.
